Research Of The Intelligent Control Methods And Applications On Two Classes Of Nonlinear Process

With regard to the two classes of non-linear process control system, this paper puts forward corresponding intelligent control methods for each other and applies them to the practice. The first type is the non-linear process control system of small time delay and fast response speed(the control period is no more than 200ms) ,the typical representative of this class system is the temperature control system for the welding process in the continuous production of welded tube(Adopting the high-frequency reaction heating way or direct resistance heating way to the welded tube ).The second type is the non-linear process control system with long time delay(the lag time is no less than 10s ), its main character is the long lag time, slow response speed, the typical representative of this class system is the temperature control system of industrial resistance heating furnace. The main work of this paper shows as follows: 1)This paper gives the overview about the intelligent control theory and its application history and current situation of the development, and points out existing problem at present and the development direction. 2)With regard to the first class of non-linear process control system, fuzzy adaptive control method with hierarchical structure has been put forward. Its designing principle is to form the whole control system with tertiary control structure: (1) Basic fuzzy control grade. In order to meet the system real-time control request, basic control grade adopts fuzzy logic control method. (2) The adaptive adjust grade. In order to meet the situation that the parameters of the controlled system are changed with time, the adaptive control method is adopted, and adjusts the fuzzy controller parameter online in timing. (3) The process state judgment grade. In order to overcome the influence that the process state changes (or be different form actual operating modes ) , and raise the robust performance of the control system, process state judgment grade adopts process state judgment as the auxiliary input, based on the system process state , uses the corresponding control parameter collection. In the paper it provides the analysis of systematic stability, and has the control simulation research. 3)With regard to the second class of non-linear process control system ,this paper introduces a new kind of associative memory neural network and its algorithm. Through adopting the factor of associative memory decays λ, it has raised the ability of identification to the non-linear system. Compared the simulation with the neural network of Elman, associative memory neural network has good identification and extensive ability. In the paper it provides the analysis of algorithm convergence property and how to get λ. 4)With regard to the second class of non-linear process control system ,this paper proposes a new fuzzy neural network structure and algorithm. In this scheme it has the three-layer fuzzy neural network controller to satisfy the early control fast speed request; In order to solve the steady-state error in the later stage, adopting the new type associative memory neural network, through the inverse identification method, as the controller to compensate the control output. Adopted the coordinated control factor β, adjusting the outputs of the fuzzy neural network and the neural network inverse controller, the controlled system is best control. In this paper it provides the convergence property analysis of the control system and how to get β. 5)Adopting the new type associative memory neural network to the multi-inputs and multi-outputs non-linear system gets the new kind of associative memory multi-inputs multi-outputs identification method; Adopting the fuzzy neural network with inverse identification structure to multi-inputs and multi-outputs non-linear system gets the new type fuzzy neural network decoupling methods for the multi-inputs and multi-outputs non-linear system. It has the identification simulation research and the decoupling simulation research to the two inputs and two outputs non-linear system.6)On the basis of carrying on experience of the theoretical research and practical experience to these two classes of non-linear process control system for many years, the author analyzes the characteristic in details of the temperature control system for the welding process in the continuous production of two-layer welded tube and the temperature control system for preheating process in the continuous production of two-layer zinc-plated welded tube and temperature control system of industrial resistance heating furnace; Adopting the fuzzy adaptive control method and neural network identification and control method to control these three systems actually has received very good application result. Computer simulation research and practical applications show that adopting the fuzzy adaptive control method and neural network identification and control method to identify and control these two classes of non-linear process control system is effective.